Image to Text (OCR)

What OCR is good at

• Printed text screenshots — emails, articles, slides
• Clean document scans — books, papers, forms
• Receipts — though OCR-only output needs manual cleanup for fields
• Subtitles / captions in screenshots

Where OCR struggles

• Handwriting — accuracy drops below 60%
• Low-resolution / blurry images — preprocess (sharpen, upscale) first
• Decorative / display fonts — Tesseract is trained on common text fonts
• Coloured backgrounds with low contrast — convert to grayscale + boost contrast first

Privacy

Tesseract.js runs entirely in your browser via WebAssembly. The image, OCR'd text, and language data all stay on your device. The language files are fetched from jsdelivr on first run (one-time per language).